EPA-600/4-77-017
April 1977
REGIONAL AIR POLLUTION STUDY
Sulfur Compounds and Particulate Size Distribution Inventory
by
Fred E. Littman
Robert W. Griscorn
Harry Wang
Air Monitoring Center
Rockwell International
Creve Coeur, MO 63141
Contract 68-02-1081
Task Order 56
Project Officer
Francis A. Schiermeier
Regional Air Pollution Study
Environmental Sciences Research Laboratory
11640 Administration Drive
Creve Coeur, MO 63141
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of trade
names or commercial products constitute endorsement or recommendation for use.
11
-------
ABSTRACT
In conjunction with the Regional Air Pollution Study being conducted
in the St. Louis Air Quality Control Region (AQCR), a methodology for
estimating the amount of sulfur trioxide (SO ) emitted by combustion sources
was developed. It is based on SO /SO ratios determined both experimentally
and from literature surveys. The most likely value appears to be 1.85% of
the SO emissions. On this basis, about 22,000 tons of SO are emitted
yearly from combustion sources.
A fine particle size inventory for the area was also developed. The
inventory gives a breakdown of particulate emissions in the range of 7 to
.01 microns, based on production rates and collection efficiencies for
point sources in the St. Louis AQCR. The information on the SO /SO
ratios and the particle size breakdown is stored in the RAPS Data Handling
System.
111
-------
CONTENTS
Abstract iii
Figures vi
Tables vii
1.0 Summary 1
2.0 Introduction 2
3.0 Sulfur Compounds 3
3.1 Scope and Definitions 3
3.2 Development of Base Data and Algorithms 3
3.2.1 Base Data 3
3.2.2 Sulfur Dioxide-Sulfur Trioxide Ratios 4
4.0 Particulate Size Inventory 12
4.1 Definitions and Scope of Inventory 12
4.2 Development of Inventory for AQCR-70 13
4.2.1 Method 13
4.2.2 Particulate Size Inventory Data Files 15
4.3 Experimental Particle Size Distribution Data 15
4.3.1 Method and Equipment 16
4.3.2 Measurements of Particle Size 16
References 26
Appendices
I. Laboratory Evaluation of the "Shell" Method of
Determination of S03 29
II. Particulate Inventory: Size - File 42
-------
FIGURES
Number Page
1 SCL Concentration Plotted Aganist 0? in Flue
Gas for Four Star Coal. 5
2 Variation of SCL Conversion to SCL with Oxygen 6
3 Percentage Conversion of SCL to SCL 9
4 Sulphur Trioxide Collector 10
5 Anderson Stack Sampler 17
6 Deposits on Stage 2 - General Motors 20
7 Deposits on Stage 4 - General Motors 20
8 Deposits on Stage 6 - General Motors 20
9 Deposits on Stage 2 - Stag Brewery 21
10 Deposits on Stage 4 - Stag Brewery 21
11 Deposits on Stage 6 - Stag Brewery 21
12 Deposits on Back-up Filter - Stag Brewery 21
13 Ammonium Sulfate Crystals on Back-Up Filter -
General Motors 23
14 Particle Size Distribution Wood River Boiler #4 25
VI
-------
TABLES
Number Page
1 Relationship Between Boiler Size and SO., Formation 6
2 Sulfur Oxide Analysis and Ratios 8
3 Particle Size Distribution Results 18
vn
-------
1.0 SUMMARY
A methodology for estimating the amount of sulfur trioxide (SO.,) emitted
by combustion sources in the St. Louis AQCR was developed. It is based on S0?/
S03 ratios determined both experimentally and from literature surveys. The most
likely value appears to be 1.85% of the S02 emissions. On this basis, about
22,000 tons of SO^ are emitted yearly from combustion sources.
An alternative method for SO., determination was evaluated and field tested.
The "Shell" method, developed originally by Goks0yr and Ross, appears to give
reliable results both in the laboratory and in the field.
A fine particle size inventory for the area was developed, based on earlier
work by MRI. The inventory gives a breakdown of particulate emissions in the range
of 7 to .01 microns, based on production rates and collection efficiencies for
point sources in the St. Louis AQCR. The information can be stored in the RAPS
Data Handling System.
Experimental data were obtained on particle size distribution of represen-
tative sources using an Andersen cascade impactor. The results indicated a
bimodal distribution peaking at around 5 microns and at less than 7 microns.
-1-
-------
2.0 INTRODUCTION
Within the framework of the Regional Air Pollution Study (RAPS) at St. Louis,
MO., a high-resolution emission inventory has been assembled. Initially, this
inventory was focused on one pollutant - sulfur dioxide - for which hourly,
measured emission data were collected. This inventory was broadened to include
all "criteria" pollutants. In addition, special inventories were also developed
for trace pollutants, heat emissions and hydrocarbons.
This study is concerned with two classes of pollutants: sulfur compounds -
primarily SO (sulfur trioxide) since a detailed S02 (sulfur dioxide) inventory
exists, and a particle size inventory, a refinement of the particle inventory
available as part of the "criteria" pollutant inventory, which does not take
particle size into consideration.
-2-
-------
3.0 SULFUR COMPOUNDS
Hourly, measured emission data for all major point sources of S0? in the St.
Louis AQCR have been gathered and are available in the RAPS emission inventory
(1-4)
data base. This work is described in a series of reports v '.
In the St. Louis area, virtually all sulfur dioxide emissions (98+%) occur
from point sources (stacks, vents, etc.). This is not to say that the remaining
emissions are unimportant, since they originate essentially at street level (auto-
motive emissions, residential and commercial heating etc.) and thus may contribute
a disproportionate share to ambient concentrations.
3.1 SCOPE AND DEFINITIONS
This report deals with SO, emissions from stationary point sources. The term
"sulfur trioxide" (SO.J is used, though it is realized that in its particulate
form, in which it is customarily collected, the compound is hydrated to sulfuric
acid (H2S04).
At stationary point sources, both SO- and SO, originate from the oxidation
of sulfur or sulfur containing compounds. The bulk of the sulfur oxides orig-
inates from the combustion of fossil fuels, while the remainder comes from pro-
cess operations such as the roasting of ores, the manufacture of sulfuric acid,
etc.
The two oxides exist side by side in an equilibrium which is largely de-
termined by operational conditions at the source. The amount of SCU present is
usually expressed as a fraction of the SOo concentration.
3.2 DEVELOPMENT OF BASE DATA AND ALGORITHMS
3.2.1 Base Data
In conformity with the National Emission Data System (NEDS) ^5', the RAPS
Emission Inventory records basic fuel consumption and process data, rather than
mass emissions of pollutants. The basic data are converted to mass flow of pol-
lutants using emission factors, stored in a separate file. The advantage of this
-3-
-------
arrangement is that it permits periodic updating of the relatively small emis-
sion factor file, without disturbing the large mass of base data.
NEDS is an annual system, based on yearly reports gathered by local or re-
gional Air Pollution Control Agencies. By contrast, the RAPS emission inventor
which covers the St. Louis Interstate Air Quality Control Region, is a collec-
tion of hourly values obtained directly for this purpose. Hourly data, based 01
a measured parameter such as fuel consumption, steam or power production are
being obtained from all the major sources of pollutants in the AQCR. A major
source for the purposes of this inventory, is one which individually emits more
than 0.1% of the total of a given "criteria" pollutant in the area. "Criteria"
pollutants for which national standards exist include SCLj NCL, CO, hydrocarbon:
and particulates.
The RAPS Emission Inventory also contains data on smaller sources, emitting
as little as 10 tons of S0? per year. Data on these sources are based on annua"
consumption or process figures, modified by an operating pattern peculiar to the
source. The pattern, which is also stored in the RAPS Data Handling System,
recor''0 the hcpirs per day and days per week for normal operation, as well as an:
holiday or vacation periods. Using this information, average hourly S02 emis-
sion values can be obtained as an output. Since these sources make up less thai
2% of all point source emissions, no significant errors are introduced by this
method.
As a result of this effort, a detailed and relatively accurate record of
SOp production exists, which can serve as a base for an SO., inventory.
3.2.2 Sulfur Dioxide - Sulfur Trioxide Ratios
In the presence of excess air in a combustion operation, a fraction of the
sulfur dioxide is converted to sulfur trioxide (SO.J according to
2 S02 + 02 -»- 2 S03 + 45.2 Kcal
The reaction is exothermic; however, the reaction rates are negligible be-
low 200°C (392°F), reach a maximum around 400°C (752°F) and taper off to zero a~
1000 C (1832 F). Rapid conversion takes place only in the presence of a catalys
-4-
-------
As would be expected from the reaction constant
K = (S02)2 x (02)
the yield increases with excess oxygen.
The information of SO, in boiler stack gases has been investigated fairly
/ r \ O
extensively. Corbett investigated the SCL formation in an oil-fired boiler.
He found that 1 to 3% of the S0~ was oxidized to SO^. The amount of SCU found
did not correlate with the percentage of sulfur in the oil or boiler conditions.
Lee ^ ' used a wet-bottom, pulverized coal-fired research boiler, several types
of coal, and varied the excess oxygen from 0.5 to 5%. He found a distinct re-
lationship on excess oxygen (Fig. 1).
V =
SO 6-0
7O 8-O
FIGURE 1: S02 CONCENTRATION PLOTTED AGAINST 02 IN FLUE GAS FOR FOUR STAR COAL
(Ref. 7)
-5-
-------
(8)
In a later study v ' Lee obtained similar results in an oil-fired furnace.
(9)
Gills v ' found a similar dependence on excess oxygen, but did not get a flat-
tening of the curve up to 12% oxygen (Fig. 2). He also found a strong de-
pendence on boiler size, with an 850 tons steam/hour boiler producing a 1% con-
version to S03 at an oxygen level of 0.5%, while smaller boilers (up to 25 tons
steam/hour) produce only a 0.25% conversion under similar conditions.
o
o
X
^
en
•+• i
61
2-0
1-0
c/i
tr
O-5
O
o
BOILER CAPACITIES
UP TO 25 tons steam/h
O 2 4 6 8 1O
OXYGEN CONTENT (% VOL)
12
Figure 2: VARIATION OF S02 CONVERSION TO S03 WITH OXYGEN (Ref. 9)
The latter relationship was confirmed by Reese ^ ', who obtained the
following results at 4% excess oxygen.
TABLE 1
RELATIONSHIP BETWEEN BOILER SIZE AND SO, FORMATION
Instal
lation size
MW
55
no
% Conversion '.
to
2
3
185 1 4
SO^
.1
.5
.4
i
i
I
.J
-6-
-------
In general, the percentage conversion falls in the range of 0.5 to 5%, with
absolute values below 50ppm of SCL*.
Our results are shown in Table 2 and Figure 3.
The concentration of S02 varied from 120 ppm for a boiler operated on
distillate oil to 2660 ppm S0? for a coal burning boiler. Average for coal burn-
ing boilers was about 1600 ppm. The SO^ concentration ranged from 2.7 to 44.3
ppm, well within the range indicated by other investigators. As indicated in
Figure 3, there appears to be a marked dependence on excess oxygen. The percent-
age of S0~ increased with increasing oxygen up to about 9%, then dropped rapidly.
This may be due to the cooling effect of large amounts of excess air. There did
not seem to be any correlation with the sulfur content of the fuel nor did there
appear to be any marked effect of boiler capacity on the amount or concentration
of SO., produced.
The RMS average S03 emission appears to be about 1.85% of the S02 emission.
This factor will be incorporated in the data handling system output program,
which will report SO^ emissions based on the corresponding S02 emissions. Using
the current figures for SCL, this amounts to an annual emission of 22,585 tons
of SO^ per year.
Analytical Methods for S03
The current standard method for SO., in stack gases is EPA Method 8 (CFR
40, 60.85, Appendix-Test Methods). In this method, the sample of stack gases
is drawn through a series of impingers. The first impinger contains 100ml of
80% iso-propanol; the second and third 100ml of 3% hydrogen peroxide. There is
a filter between the first and second impinger to retain entrained particulates.
The contents of the impingers are analyzed for sulfate using the barium
perchlorate-thorin method.
(9)
* An interesting exception was found by Gills v ' in brick kilns, where up to
28% of the sulfur oxides were in the form of SO.
-7-
-------
CO
o
CM
CO
CO
I—I
CO
>-
_l
team/hr
_Q
O
13
E
Q.
O
O
X
LJ_
co
to
_Q
E
CL
Q.
n
0
X
u_
t_3
cn
CM
O
0
0
O
Ln
*d-
co
Cn
r-
CM
CO
CM
r^
CM
CD
i-D
Ln
CD
CM
s
CM
=»=
CU
QL
O
o
3
CM
CO
O
0
O
O
O
CO
^.
i2
o
Ln
CO
Ln
CO
co
CM
Ln
cn
CO
CO
CM
CO
O
o
o
o
CO
un
un
cn
^
±
un
CO
CM
CO
co
CM
VO
^o
CM
0
^
"O
na
-C
cn
•i—
LO
CM
CO
O
O
o
o
CM
CM
r-
i —
un
r--
ID
r-
UD
O
CTl
r—
cn
to
CO
CO
o
o
0
0
CO
cn
co
CM
CM
CO
co
CM
CO
5
o
cn
CO
CTi
ID
LT>
cn
o
cn
CO
CJ3
o
0
CO
*
o
0
0
o
o
CM
LO
o
r-
CO
^
to
un
CM
CO
o
o
to
CM
0
o
o
i.
co
r^
r-
cn
CM
Ln
CO
CO
CO
CO
CM
oo
CO
0
CM
| Average
ID
CO
CU
cn
(O
O)
CO
&L
O)
"O C7)
Ol i.
-£Z CU
cn >
•.- ro
CU
-8-
-------
°7o S03
3-1
2 _J
0
2 4
8 10 12 °7o 02
FIGURE 3
PERCENTAGE CONVERSION OF S02 TO S03
-9-
-------
Recent work cast doubts on both accuracy and reproducibility of Method
8^ . The method assumes that only SO, (sulfuric acid mist) will be retained
in the first impinger and the filter (both of which are analyzed together).
(12)
However, Hillenbrand^ ' found that substantial amounts of SCL are retained in
the first impinger, some of which is subsequently oxidized to SO.,, thus con-
tributing to high results. For this reason a different technique was used,
(13)
which was first described by Goks0yr and Rossv and subsequently verified by
(14)
Lisle and Sensenbaugh . The method is generally referred to as the "Shell"
method, as it was developed in their laboratories. The method is based on the
condensation of sulfuric acid mist at temperatures below its dew point (but
above the dew point of water) in a condenser backed up by a fritted glass fil-
ter (Fig. 4). The condensate is washed out and titrated.
STOPPER
GRADE 4
S.MERED
GLASS
DISC
FIGURE 4: SULPHUR TRIOXIDE COLLECTOR (Ref. 12)
-10-
-------
Data presented in references 12 and 13 indicate that adsorption of S03
is essentially complete, repeatability is excellent, S0? in concentrations
4. ^
up to 2000ppm does not interfere and a precision of - O.Sppm of SCL can be
readily attained.
The method was then evaluated in our laboratories. The results of the
evaluation are shown in Appendix I; they indicate an average 100.1 - 6.5%
recovery with no significant interference from any of the variables tested.
-11-
-------
4.0 PARTICULATE SIZE INVENTORY
Emissions of participate materials constitute a more complex problem
than gaseous emissions, since the properties of particles are determined
not only by their composition, but also by their size and shape. In fact,
the most important properties of particles, their effect on visibility,
their life-time as suspended materials, and to a large extent, their effect
on health, are all determined by particle size. On all of these counts small
particles, 5 microns or less in diameter, are responsible for most of the ob-
served effects.
The common methods of collection and reporting of particulate emissions
do not distinguish particle size. Total particulate matter is reported on a
weight basis, which biases the results in favor of large particles. Since large
particles are only of local importance - they settle out rapidly - and are gen-
erally not involved in health effects because they are readily retained by the
body's screening mechanisms, there are good reasons why particulate emission
data should be reported in such a way as to provide maximum information on small
particles.
4.1 DEFINITIONS AND SCOPE OF INVENTORY
There is no universally accepted definition of "fine particles", but most
authors agree on a range of 3 to 5 microns as the upper limit. Particles small-
er than approximately 5 microns have settling velocities in still air of the
order of 0.01 cm/sec and tend to stay aloft almost indefinitely. Particles can
be either solid or liquid.
The most up-to-date study of fine particulate emissions is contained in EPA
Technical Report entitled "Fine Particulate Emission Inventory and Control
Survey" ^ . The methodology contained in that report was applied to the St.
Louis AQCR. In addition, samples were taken at representative emission sources
using an Andersen cascade impactor. Data developed from this study are also
included.
-12-
-------
4.2 DEVELOPMENT OF INVENTORY FOR AQCR-70
In order to prepare a particle size inventory within the scope of RAPS,
compatible with the NEDS and RAPS Data Handling Systems, the procedure outlined
below was used. No effort was devoted to the inventory of area sources, mobile
sources, chemical and physical characterization of these particulates.
4.2.1 Method
The method put forth in the "Fine Particulate Emission Inventory and Control
Survey" uses the following equation for the calculation of particulate emissions:
r PefCf n
drd2 = *
where
d-j-d^ = emission rate for particles with diameter between d, and d~
P = production rate
ef = emission factor (uncontrolled)
C, = percentage of production capacity on which control equipment is in-
stalled (for that device)
f-j(d) = emitted particle size distribution
f~(d) = penetration = (1 -fractional efficiency of control system)
The size ranges are (in microns):
.01 - .05, .05- .1, .1 - .5, .5-1, 1 -3, 3-7.
The data sources are:
(1) RAPS coding forms (or NEDS computer listing)
(2) "NEDS Source Classification Codes and Emission Factor Listing" (SCC
listing), July 1974.
(3) "Fine Particulate Emission Inventory and Control Survey" (EPA-450/3-
74-040), January 1974.
The equations used for calculating the total particulate emissions are:
E, , (total) = E, . (controlled) E, , (uncontrolled) (2)
Q-i ~UQ UT~UO ' Q -i ~" Q^
-13-
-------
where E^ _^ (controlled) is expressed in equation 1, and
Pe (1-Ct}
E, _, (uncontrolled) = f * f (d]
Q] °2 2000 rl (Q)
The assumption used here is that f~ (d) applies to that fraction of the
emission which is specified by C. and 0-C.) has no control and therefore
f2(d) = I-
The algorithm for a computer program may be something like the following:
I For every point source
la Look up from RAPS coding form, the SCC code Card 4
Ib Look up P (annual data) Card 5
Ic Look up C. (control efficiency) Card 3
Id Look up CID (control device ID code) Card 3
II From EFACTR file. Look up e,. (uncontrolled emission factor) for the
corresponding SCC number.
Ill From SIZE file. Look up size distribution in fractional values for each
size range for the corresponding SCC number.
IV From the EFCNCY file. Look up the fractional efficiency of each size
for the corresponding control device as identified by the code number
CID.
V Calculate the emissions using equations 1, 2 and 3.
The following are examples of the three computer input data files:
File Name: SIZE
SCC Code .01-.05 .05-.1 .1-.5 .5-7 1-3 3-7
Fractional Values
File Name: EFACTR
SCC Code Emission Factor
Pounds per Ton
-14-
-------
File Name: EFCNCY
CID .01-.05 .05-.1 .1-.5 .5-.1 1-3 3-7
Efficiency
4.2.2 Particulate Size Inventory Data Files
Of the three files required for calculations of the particulate size inven-
tory, one, the emission factor file EFACTR, is already contained in the RAPS in-
ventory. The other two were developed on this Task and are given in Appendix II.
The SIZE-file, in a matrix form, gives the particle-size distribution of
emitted particulates for any one of the forty-four SCC-codes listed in column 2.
Each column between columns 3 through 8 lists the fractional value of the total
particulate effluent that falls within the corresponding particle-size range.
All values to the right of the double line (columns 2-8) are keypunched for
computer input with the READ format: (18, 6F4.0), blanks = 0. The fractional
values in the F-format are left justified with no decimal points.
The EFCNCY-file lists the fractional efficiency of the effluent control de-
vice for each particle-size range. The control device is identified by the CID
number under column 2 and the particle-size ranges have the same diameter group-
ings as that in the SIZE-file.
All values to the right of the double line (columns 2-8) are keypunched for
computer input with the READ format:
(13, 6F4.0), blanks = 0
The fractional values in the F-format are left justified with no decimal points.
Both files were keypunched. The cards are available for input into the RAPS
Data Handling System.
4.3 EXPERIMENTAL PARTICLE SIZE DISTRIBUTION DATA
In connection with the emission factor verification program carried out as
part of the RAPS study, data were gathered on particles size distribution of a
number of representative sources.
-15-
-------
4.3.1 Method and Equipment
Particle size testing was performed with an Andersen Stack Sampling head
coupled with the apparatus used for standard EPA method for particulates. The
Andersen is a fractionating inertial impactor which separates particles accord-
ing to aerodynamic characteristics.
The Mark II sampling head consists of a stainless case, plate holder and
nine jet plates. The plates have a pattern of precision-drilled orifices. The
nine plates, separated by 2.5 millimeter stainless steel spacers, divide the
sample into eight fractions or particle size ranges. The jets on each plate are
arranged in concentric circles which are offset on each succeeding plate. The
size of the orifices is the same on a given plate, but is smaller for each
succeeding downstream plate. Therefore, as the sample is drawn through the
sampler at a constant flow rate, the jets of air flowing through any particular
plate direct the particulates toward the collection area on the downstream plate
directly below the circles of jets on the plate above. Since the jet diameters
decrease from plate to plate, the velocities increase such that whenever the
velocity imparted to a particle is sufficiently great, its inertia will overcome
the aerodynamic drag of the turning airstream and the particle will be impacted
on the collection surface.
The Mark III is identical to the Mark II except the location of the orifices
in the plates have been modified to permit the use of a special collection sub-
strate (glass fiber in our tests). This permits lighter tare of weights for
gravimetric analyses and a collection of material for chemical analysis. Figure
5 illustrates the Andersen sampling head and an exploded view of the plate holder
and plates.
4.3.2 Measurements of Particle Size
Particle Size Distribution measurements have been conducted at five of the
seven test sites sampled in 1975. Initially only the Andersen Mark II plates
were available. Because of this the only results available at the first test
site are the weight distribution. On subsequent tests, runs were made with both
the Mark II plates and Mark III plates with glass fiber filters for comparison.
Sites that have been tested for particle size are shown in Table 3. Some of the
filter samples were inspected microscopically and a few of these were also
-16-
-------
AIR FLOW
0
FIGURE 5
ANDERSEN STACK SAMPLER
-17-
-------
analyzed by x-ray fluorescence. A summary of the results of the testing is given
in Table 3. Particle size is given as aerodynamic size for spherical particles
with unit density.
TABLE 3
PARTICLE SIZE DISTRIBUTION RESULTS
Source
111. Power - Wood River
Highland Electric
Stag Brewery
General Motors
Amoco
Sec Code
1-01-002-02
1-01-002-08
1-02-002-05
1-02-002-09
3-06-001-02
3-06-001-03
>7y
22.5
26.6
37.4
14.3
13.9
% vs
3-7y
22.8
18.9
16.0
24.4
8.9
Parti cl
1-3U
18.5
10.0
7.6
18.5
22.0
e Size
0.5-lvi
8.3
12.7
18.3
9.2
18.8
< 0.5u
27.9
31.8
20.7
33.6
36.4
At General Motors, fourteen tests were performed to evaluate variations of
testing methods consisting of placing the Andersen in-stack, out of stack (in oven)
using Mark II plates and Mark III plates with filters. Each of these methods has
its advantages which may make it desirable for any one individual test. The main
objective of these tests was to arrive at a testing arrangement to be used on all
subsequent tests. As it turned out there was no clearcut single method which
proved better than the others.
Sampling in the stack avoids any problems with extracting a sample and hav-
ing some of it deposited in the probe. Also the sample head is at the same temper-
ature as the stack gases which avoids any problems of condensation. In-stack samp-
ling, however, means the impaction surface is vertical and is subject to having
the sample dislodged in handling. When sampling must be done vertically in a duct,
from the top down, this method cannot be used.
Sampling with the Andersen sampler in the sample oven at the end of a heated
probe affords much better handling. The sample head can be kept vertically with
the plates horizontal at all times. The sample head is also clamped in place and
doesn't have to be threaded on to the probe, which avoids more handling.
Isokinetic sampling rates can be determined more readily when the Andersen
is in the oven since the probe has a pi tot attached and the probe remains in the
stack (for in-stack sampling a pi tot measurement is made, the pi tot is removed
-18-
-------
and the sampler is inserted to approximately the same position). There are two
problems with sampling this way: the oven can be heated only to 350°F, which
may not be as high as the temperature in the stack, and larger particles tend
to be deposited in the probe, which lowers the weight of the deposit on the first
two plates.
Parallel sampling with both the Mark II plates above and the Mark III plates
with filters indicates that there isn't any significant difference in the weight
of catch and the size distribution between these two methods. If the Mark II
model is used, the number of tests is limited by how many sets of plates are avail-
able. With the Mark III plates and filters more runs can be performed by chang-
ing the filters between runs with the available time being the only constraint
on the number of runs. More care must be taken in assembling the Mark III, since
the filters are pre-cut to match the plates and must be properly aligned to avoid
blocking any holes.
As a result of these comparison tests, it was decided that testing would be
performed with the Mark III plates and filters and that the Andersen sample head
would be placed in the oven for ease in handling and subsequent analysis.
Photomicrographs have been made by Illinois Institute of Technology Research
Institute (IITRI) of samples collected on each stage from three Andersen runs.
These pictures confirm that the Andersen does in fact separate by particle size
as the instructions would indicate. Evidence of this is shown in Figures 6, 7
and 8 from General Motors and Figures 9, 10, 11 and 12 from the Stag Brewery.
Figure 6 is from stage 2 taken at 163x. This shows a high percentage of
fly ash and partially fused clays and minerals, average particle size is approxi-
mately 6 microns. Figure 7 is from stage 4 taken at 163x. This shows much small-
er particles, a high percentage of fly ash and more Fe^CL, and an average
particle size of approximately 2 microns. Figure 8 is from stage 6 taken at 163x.
This shows mostly submicron partially burned coal, fly ash and Fe20o-
For spherical particles with unit density stage 2 should have separated
from 10.9 to 17 microns, stage 4 from 5.0 to 7.3 microns, and stage 6 from 1.7 to
3.2 microns. Since fly ash has a density between 2 and 3, these stages will
actually separate smaller particles.
-19-
-------
FIGURE 6
DEPOSITS ON STAGES
2,4 AND 6
GENERAL MOTORS
FIGURE 8
-20-
-------
-21-
-------
Figure 9 from Stag Brewery is from stage 2 taken at 406x. This shows lots
of incompletely combusted coal, partially fused glassy material and Fe^O,
partially fused and coating other particles. Average particle size is approxi-
mately 5 microns. Figure 10 is from stage 4 taken at 406x. This shows fine fly
ash spheres most of which are dark due to iron in solid solution and some miner-
als and fine carbonaceous particles. Average particle size is approximately 2
microns. Figure 11 is from stage 6 taken at 406x. This shows what appears to be
black carbonaceous material which hit as a liquid or is particles suspended in a
liquid. There is very little fly ash or else it is below 0.5 micron. Figure 12
is from the backup filter taken at 406x. This shows extremely fine liquid drop-
lets with trapped fine carbonaceous particles and extremely fine sulfate par-
ticles.
Microscopic analysis of the filters has indicated that sulfate crystals form
on the filters in increasing amounts on descending stages to the point where the
backup filter sample is mostly sulfate. Personnel from IITRI have indicated that
these crystals are ammonium sulfate and that they have grown on the filters.
Figure 13 is from a backup filter from a test at General Motors, taken at 163x.
Clearly, these crystals could not have passed through the Andersen impactor.
The mechanism for the formation of these crystals is still not understood.
Apparently, there is a reaction between ammonia in the flue gases with sulfuric
acid on the filters. To check that this reaction didn't take place from ex-
posure sometime later, one backup filter was sealed in an air-tight enclosure at
the test site and then examined immediately after opening the sample container.
This sample also showed a large amount of crystals.
A few of these backup filters have been analyzed for acidity. Approximately
17% of the amount of sulfuric acid measured in the stack at General Motors.was
found to be entrained by the backup filter and by the total particulate filter
on an EPA particulate run. Whether this is due to condensation and entrainment,
or a gas-solid phase reaction, is not known. At these temperatures, 440°F in-
stack and 350°F in the oven, sulfuric acid vapor should not condense.
One test run indicated that temperature has some relationship to the amount
of material in the backup filter. Two identical Andersen runs were made at
General Motors with the sample head in the oven. The first test was with an oven
-22-
-------
FIGURE 13
AMMONIUM SULFATE CRYSTALS
ON BACK-UP FILTER
GENERAL MOTORS
-23-
-------
temperature of 300°F and the second with a temperature of 370°F. While the first
8 stages were very similar in weight, there was twice as much material collected
on the backup filter in the first test than in the second.
The particle size distribution from all of the tests performed to date shows
a bimodal distribution, generally with a peak around stages 4 or 5 and a large
peak on the last backup stage. A typical curve is shown in Figure 14. The large
amount of sulfate crystals on the backup indicates that perhaps 30% of that amount
is sulfuric acid and should not be included. But even after this is subtracted
there are two peaks, one around 5 microns and the other less than 0.7 micron.
-24-
-------
45-
40-
35-
30-
25-
20-
15-
10-
5-
FIGURE 14
PARTICLE SIZE DISTRIBUTION
WOOD RIVER
BOILER #4
I
11 10 987654 32 1 0
ECD, microns
-25-
-------
5.0 REFERENCES
1. Llttman, F. E., "Regional Air Pollution Study Point Source Methodology and
Inventory" Rockwell International, EPA 450/3-74-054, October 1974
2. Littman, F. E. and R. W. Griscom "RAPS Point Source Emission Inventory - Phase
II" Air Monitoring Center, Rockwell International, EPA Contract No. 68-02-
1081, July 1975.
3. Littman, F. E., R. W. Griscom and Otto Klein "RAPS Point Source Emission Inven-
tory" Air Monitoring Center, Rockwell International, EPA Contract No. 68-02-
1081 , February 1976.
4. Pierre, John, and W. Tillman "RAPS Point Source Emission Inventory Data Hand-
ling System" Air Monitoring Center, Rockwell International, EPA Contract No.
68-02-1081 , February 1976.
5. "Guide for Compiling a Comprehensive Emission Inventory", U. S. Environmental
Protection Agency APTD-1135, March 1973.
6. Corbett, P. F., "The SOo Content of the Combustion Gases from an Oil-fired
Water-tube Boiler", J. Inst. Fuel, Aug. 1953, p. 92.
7. Lee, G. K. et al, "Effect of Fuel Characteristics and Excess Combustion Oil on
Sulfuric Acid Formation in a Pulverized-coal-fired Boiler", J. Inst. Fuel,
Sept. 1967, p. 397.
8. Lee, G. K., et al , "Control of S03 in Low-pressure Boiler", J. Inst. Fuel,
Feb. 1969, p. 67.
9. Gills, B. G., "Production and Emission of Solids, SO, and NO,, from Liquid Fuel
Flames, J. Inst. Fuel, Feb. 1973, p. 71.
10. Reese, J. T., et al, "Prevention of Residual Oil Combustion Problems By Use
of Low Excess Air", Trans. ASME, J. Engrg. Power, 1965, V87A, p. 229.
11. Hamil, H. F., et al, "Collaborative Study of EPA Method 8 (Determination of
Sulfuric Acid Mist and Sulfur Dioxide Emissions from Stationary Sources)",
EPA 650/4-75-003.
-26-
-------
12. Hillenbrand, et al, "Chemical Composition of Participate Air Pollutants from
Fossil-Fuel Combustion Sources", Battelle Columbus Labs, March 1973, EPA-
R2-73-216, PB219.009.
13. Goks0yr, H., and K. Ross, "Determination of Sulphur Trioxide in Flue Gases",
J. Inst. Fuel V35, p. 177 (1962).
14. Lisle, E. S. and J. D. Sensenbaugh, "Determination of Sulfur Trioxide and Acid
Dew Point in Flue Gases", Combustion 36_, 12, (1965).
15. Weast, T. E., et al, "Fine Particulate Emission Inventory and Control Survey",
Midwest Research Institute, EPA 458/3-74-040, Jan. 1974.
-27-
-------
APPENDIX I
-28-
-------
LABORATORY EVALUATION
OF THE "SHELL" METHOD
OF
DETERMINATION OF SOg
The "Shell" method for determination of sulfur trioxide (sulfuric acid)
in flue gas is based on its selective condensation from the flue gas. This
is achieved by utilizing the relatively high (60-90°C) dew point of SO^. At
this temperature only the sulfuric acid condenses from the flue gas and there-
fore it can be determined rather easily.
Flue gas is drawn through the condenser at a rate of 2 liters per minute
for 10-20 minutes depending upon the SO, level in the flue gas. Particulates
are removed from the flue gas sample by using a plug of glass wool as a filter.
At the end of the sampling period, the H^SO* is washed out of the condenser
with 5% solution of isopropyl alcohol in water. The combined washings were
titrated with 0.02 N NaOH using bromophenol blue as indicator.
The laboratory evaluation of this method had a dual purpose. The first
was to check the accuracy of the method under the experimental conditions and
secondly, to determine which of the experimental parameters may affect the per-
formance of this method. For the latter, stack conditions had to be simulated
in a way which would allow adjustment of each parameter to predetermined levels,
The accuracy of the method was tested by duplicating the experimental work of
tio
(2)
E.S. Lisle and J.D. Sensenbaugtv . The effect of the experimental conditions
on the accuracy of the method was evaluated by using the Plackett-Burman
statistical design of screening process variables. This method is based on
balanced incomplete blocks. A good example of applying this method to a chem-
(3}
ical process has been published by R.A. Stowe and R.P. Mayerv '. With this
method it is possible to effectively screen all the experimental parameters and
to find out which of them most likely will affect the overall process, by per-
forming only a small fraction of experimental work usually required for other
methods of screening variables. For example, a complete factorial design for
fifteen variables at two levels requires 32,768 experiments; with the Plackett-
-29-
-------
Burman method the same number of variables can be screened effectively with
(3)
only 16 experiments '. It should be emphasized, however, that this method
does not optimize the process; it only indicates which of the parameters do
not affect the process.
EXPERIMENTAL
The experimental set-up used in this study is given in Figure 1. A
special condenser thermostated at 60-90°C was used for the collection of the
condensed H^SO^. The simulated flue gas is introduced at the end of the con-
denser which consists of a spiral glass tube followed by a coarse glass frit-
ted disc. Both the spiral and the glass fritted disc are kept at constant tem-
perature (60-90°C) by circulating water from a heating bath. The H^SO, gener-
ator consists of a quartz tubing heated electrically by a spiral of nichrome
wire insulated by several layers of asbestos tape. With this arrangement the
temperature of the hLSCL generated can be adjusted at the desired level and kept
constant within 10°F. Dilute sulfuric acid solution is added at a constant rate
by a peristaltic pump through a hypodermic needle and serum cap in the top open-
ing of the HUSO* generator. The rate of ^SO, addition can be altered by using
pump tubes of different diameter. The flow rate of the gases (O^, N2, S02) was
adjusted and maintained at the proper levels with a combination of valves and
rotometers. The total flow was checked by a rotometer at the outlet of the
condenser.
PROCEDURE
The HUSO* generator was calibrated by titrating the amount of acid delivered
by the peristaltic pump at the upper end of the generator for a certain period
of time (about 10 minutes) for the two pump tubes and the two H?SO» solutions used
throughout the experimental work. The nominal flow rates of the pump tubes used
were 0.42 and 0.70 cc/min and the normality of the h^SO, solutions was 0.01 and
0.03 N. Tables 1, 2, 3 and 4 give the calibration of the HUSO, generator for the
above flow rates and the sulfuric acid solutions. The results are expressed in
u. equiv/min. The actual experiments were conducted in a similar manner. Sul-
furic acid solution was delivered to the HUSO, generator by the pump for about
ten minutes and collected in the condenser. The condensed hLSO* was washed out
of the condenser with 5% isopropyl alcohol in water and the combined washings
-30-
-------
13
Q_
o
-M _
ra
S-
OJ
D_
CM
o
O
o
O
CM
UJ
uj o:
Oi CO
ID t—i
CD _i
-31-
-------
Table 1
CALIBRATION OF THE S03 GENERATOR
Run #
1
2
3
4
5
Time
Sec.
601.
599.
599.
600.
600.
Nominal Pump Rate =0.42
Normality of H^SO, Solution
cc of 0.02 N NaOH
7 2.58
7 2.42
5 2.45
3 2.30
4 2.32
Average
Table 2
cc/min
= 0.01 N
y equil/min
5.14
4.84
4.90
4.60
4.63
4.82 2: 0.22 y. equiv/min
CALIBRATION OF THE SO. GENERATOR
o
Run #
1
2
3
4
Time
Sec.
501 .
601 .
600.
600.
Nominal Pump Rate = 0.42
Normality of H2S04 Solution
ml of 0.2 N NaOH
0 7.46
5 7.82
4 7.52
5 7.66
cc/min
= 0.03 N
y equil/min
14.89
15.60
15.03
15.30
Average 15.20 +_ 0.31 y. equiv/min
-32-
-------
TABLE 3
CALIBRATION OF THE S03 GENERATOR
Nominal Pump Rate = 0.7U cc/min
Normality of H2S04 Solution = 0.01 N
Time Titrant
in #
1
2
3
4
5
Sec.
630.6
600.0
689.8
600.7
600.4
ml of 0.02 N NaOH
4.60
4.46
4.98
4.44
4.42
y equil/min
8.15
8.92
8.66
8.87
8.83
Average 8.81 t 0.10 y.equiv/min
TABLE 4
CALIBRATION OF THE S03 GENERATOR
Nominal Pump Rate = 0.70 cc/min
Normality of H2S04 Solution = 0.03 N
in #
1
2
3
4
5
Time
Sec.
599.9
600.2
599.2
600.7
602.2
Titrant
cc of 0.02 N NaOH
12.80
12.23
12.26
12.42
13.24
y equil/min
25.60
24.45
24.55
24.81
26.38
Average 25.16 t 0.82 y.equiv/min
-33-
-------
were titrated with 0.02 N_ NaOH. Throughout this work all the experimental
parameters were varied at two levels: one high level and one low level des-
ignated here as (+) or (-) respectively. Table 5 gives all the experimental
parameters examined in this study and their respective high and low values.
The resulting efficiency of collection of the generated H?SO» vapors was
determined by dividing the recovered amount of H?S(L by the amount of SO^ de-
livered into the system (Tables 1 and 4).
RESULTS AND DISCUSSION
As it was mentioned previously, the purpose of this study was to first de-
termine the efficiency of the system under the recommended conditions and sec-
ondly to screen all the experimental parameters and determine which of them af-
fect the efficiency of the system.
Table 6 summarizes the results obtained by using the system under the rec-
ommended conditions. No SCL was used in these experiments because the main pur-
pose was to determine the efficiency of collection of H9SO,, from flue gas. These
(1)
experiments were performed in the manner recommended by Lisle and Sensenbauglv .
The samples were introduced in the evaporator by a syringe through the serum cap
without the use of the proportioning pump. The average recovery was found to be
equal to 100.1 - 6.5%. It should be noted, however, that no extra effort was
made to optimize any of the experimental conditions and therefore these results
represent data obtained by a casual application of this method. A close inspec-
tion of the results tabulated in Table 6 shows that sources of positive (recov-
eries > 100%) and negative (recoveries < 100%) errors do exist and therefore an
examination of the parameters affecting the accuracy of the method appeared to
be necessary. The parameters listed in Table 5 were tested by the method of
Plackett and Burman by using a matrix for sixteen runs and fifteen variables.
Figure 2 gives the Plackett-Burman matrix used in this study. Five out of the
fifteen variables were blank "dummy" tests from which the standard error of the
method was calculated.
The statistical analysis of the results is given in Table 7. In this table,
confidence levels are shown only to the 70% level; the remaining variables are
considered to have an insignificant effect on the method within the studied ranges,
Therefore from the ten variables studied only four may have an effect on the ac-
-34-
-------
TABLE 5
VARIABLES CHOSEN FOR STUDY
Code letter of Levels
the variable Variable Low(-) High(+)
A Temperature of Condenser (°C) 60 i 3 90-3
B Temperature of Evaporator (°F) 550 650
C 02/S02 Ratio 149 223
(D) Dummy — —
E Total Flow (liter/min) 2 4
F Flow Rate of H2S04 Solution (cc/min) 0.42 0.70
(G) Dummy — —
H Elapsed Time Prior to Rinsing the 1 10
Condenser (min)
I Volume of Solvent for Each Washing (ml) 10 25
J Total Volume of Solvent Used for Each 135 185
Experiment (ml)
K Size of Hypodermic Needle (gauge) 26 20
(L) Dummy — —
M Normality of H2S04 Solution 0.01 0.03
(N) Dummy — —
(0) Dummy — —
-35-
-------
TABLE 6
COLLECTION EFFICIENCY OF THE METHOD
OF S03 DETERMINATION IN FLUE GASES
Volume of acid = 4.0 mil; total flow = 3.96 liter/min; nitrogen to oxygen ratio
•e 85°C; temporal
m. equiv. H?S04
20:1; condenser's temperature 85°C; temperature of H^SO, generator 600°F
Run #
1
2
3
4
5
6
7
8
9
10
11
12
Taken
0.040
0.040
0.040
0.040
0.080
0.080
0.080
0.080
0.120
0.120
0.120
0.120
Found
0.039
0.042
0.041
0.047
0.077
0.077
0.074
0.077
0.123
0.119
0.120
0.115
% Recovet
97.50
105.00
102.50
117.50
96.25
96.25
92.50
96.25
102.50
99.17
100.00
95.83
Average % Recovery 100.0 +_ 6.5
-36-
-------
PLACKETT-BURMAN MATRIX FOR DETERMINING
THE EFFECTS OF FIFTEEN VARIABLES
AT TWO LEVELS USING SIXTEEN RUNS
- = LOW
Random Run Van'able + = "^
Number Order A B C (D) E F (G) H I J K (L) M (N) (0) % Recovery
1 1 + + + + _+_+ + __ + __ _ 92i4
2 2 + + +_+_ + +_-+--_ + 93i9
3 3 + + _ + _+ + _- + ___+ + 92.5
4 8 + _+_++_- + -__ + -{- + 94 j
5 4 - + _ + +_-+__- + + + + 87A
6 6 +-+ + -_ + ___+ + + + _ 86.6
7 9 _ + +_-+___ + + + + - + 83J
8 12 + + _-+_-_ + + + + -+ - 109.3
9 10 +__ + _-_+ + + +- + _ + 87.3
10 13 __+_-_ + + + +- + _+ + 102.1
11 14 _ + ___+ + + + -+- + + - 83.7
12 16 +___++ + +_ + - + + _ - 95.4
13 5 ___. + ++ + _ + _+ + _-+ 98.5
14 7 __+ + ++_+- + +__+ _ 107.5
15 11 -4--H4-+- + - + +-- + - - 81.8
16 15 ____-_-----_-- - 84i3
FIGURE 2
-37-
-------
OO
cc
LU
1
r~~
LU
SI
c£
Q-
1 1
o
i—
CO
J — 1
""
1
^
LU
_J
<
QJ
CJ
1
CD
CD U
O C
C CD 1 CD 1 ! LD
ro ~o i co i i cn
O -i—
..— 4-
4- E
•r- O
E CJ)
CO
CO
CD
>
'+J
ra
"CD -x
^^ j_i
co co CM cn "=3-
CD " — CO ft- <-O
I— o «tf- a <*•
i • • • •
+J r— r— O CM
1
4^ cn i — ^d~ cn o^
O O CO CM i — «3" Cn
CM "3- O i — CO
cj_ O O O O O
LU + O O O O O
- — 1 I
S- O
CD -f-3
to ra
C i-
CD CD O
£ -O Q.
ro E ro o
2: o > T-
O LU -t-> 3
ra o
4- 4~ C£ r-
o o u_
CM
tU • O >, r—
__ Q- O- OO E ra
r- « -^ ^ , j
_Q £ £= ^- != -I-3
(rt CD CD CM rj O
.r- 1— 1— 0 Q t—
^,
ra
>-
CD
0 <£ CO CJ Q LU
1 ! 1
I i 1
1^. O
co vo
ro CO
0 0
o co •^r
i — ro •=d-
r— 1— CM
0 O 0
000
1 1
CO
c
•1 —
CO
cz
• 'I —
E rv*
1 —
0 0
CO +->
•* S-
0 0
00 -i—
CM S_
re a.
4- CD
O E
• J —
CD 1—
ra "O
ry rjj
>j co
2 E D-
0 E rO
U_ Q LU
u_ cn rn
1
1
*3-
co
o
CM
0
0
cn
•r—
_d
CO
ra
~r^
c-
(^ >
ra
LU
c
o
4 —
-4-J
E
aj
o
OO
4-
o
cu
5
o
k— 1
LO
co
0
1^
•-"
o
LO
o
o
B
-t-J
cx
X
LU
_E
(J
ra
LU
s_
o
4-
4_)
E
CD
>
o
OO
4-
o
a
^:
rs
o
ro
0
h-
•-3
i i cn i t
i i cn i i
cn co
00 CO
CD CD
^i- t — cn i — i —
CM
CM i — cn u~> o
o o o o o
o o o o o
1 I
o
CD ••-
1 — 4_J
-a rs
CD i—
CD O
-zz oo
CJ <*•
•r- O
E oo
i- CM
cu re
-a
O 4-
Ck- O
-E >>
-M
4- -r-
O i—
^} ra ^> ^*s
CD E E E E
N E s- e E
oo Q rr a Q
?*£ ~j 's: ^z o
. ,
1
^ — •"
~z~
o
r—
OJ
cn
ra
s_
CD
ra
CD
_E
-l->
CO
<~
E
~i'
_E
CT:
"r~
-E
QJ
cn
£
CO
rO
CD
4-3
4-
o
CD
o
CD
s_
cu
tj_
•p—
•a
QJ
_[ 1
CO
S-
CD
CD
E
ra
a.
_E
u
rO
QJ
o
+J
o
CD
4-
4-
c
s.
c
U.
c
f
c
c
c
I/
Q
5.
CO
CO
<:
J
•)
3
J
>
,)
J
J
J
_)
2_
3
J
)
J
5
3
s_
5
J
•'-
4-
cu
•
1 ' 1
oo
cJ
*-
ra
•^
_
CD
JD
rO
il
ra
>
4-
o
u
CD
4-
4-
cu
CD
J=
t/>
LU
CD
CD
JE
3
«
4 —
4-
CD
•
> LU
LU •
00
II
-*-•
^^
-^-*
CD
^*
•r—
CT
CO
•1 —
CD
r~~
ra
si
ra
>
ro
4-
o
QJ
Z3
ra
^"
t/)
O)
1
I
•X
•X
QJ
-E
•*->
4-
o
CJ
_Q
ci:
--•'
*—
CD
-~
-*-*
E
E
ra
^J
0
CD
4-
4-
o
^
E
-a
CL)
_E
CO
LU
s
_r-
^3
CM
T3
LU
1X1
' .
(
II
4—
4~
o
•
! i 1
•
OO
^>
-O
E
CD
>
cn
u
o
4—
QJ
s_
o
S-
QJ
~c3
>-
-^
E
fo
^
»iM,
««"
a
1 —
j:
It
-p.
s.
f^
^
F
1^
-c
-38-
-------
curacy of the method and should be studied further for the optimization of the
total system. These four variables are the temperature of the evaporator (B),
the total flow (E), the total volume of washing solution (J) and the normality
of the HpSO* solution used (M). It should be noted at this point that from these
four variables, the two (B and M) are very closely related with the experimental
conditions used for generating simulated flue gas in the laboratory and therefore
may not be associated with the application of the method in the determination of
SO^ in real flue gas. The other two (E and J) are associated with the method and
appear to be the most significant parameters which may affect the accuracy of the
SOo determination in flue gas. The total flow (parameter E) most likely affects
the condensation of S03 from the flue gas and the total volume of washing solu-
tion (parameter J) is related with the efficient washing of the condensed H2S(L.
These two parameters are most likely the ones on which proper attention should be
given in the application of this method for determination of SOo in flue gas.
-39-
-------
REFERENCES
1. E.S. Lisle and J.D. Sensenbaugh, Combustion 36., 12 (1965).
2. R.L. Plackett and J.P. Burman, Biometrica 33_, 305 (1946).
3. R.A. Stowe and R.P. Mayer, Ind. and Eng. Chem. 58., 36 (1966)
-40-
-------
APPENDIX II
-41-
-------
ro
un
o
i
o
o o
o
LO
ro
o
o
o
o
o
LO LO
o o
LU
1 — 1
LJ_
1
LU
M
1 — 1
oo
. .
>—
o;
o
1—
LU
>
1
LO
LU O
CD
^
0
o
o
•
Di
LU 3-
NJ LO
h-H •
OO 1
, —
LU
1
en
CD
0
0
1 —
et
1
o
1— ( ^-
1— I—
CsL \
I—I
CtL
D_
LO
LO
OO
ct
oo
D_
LU
D-
O
O
OO
to
OJ
to _J
d) -M to
>, C C
S- d) r—
+J
r—
03
Q.
to
rc
J^
Q-
+->
C
O)
a>
o
CXI
o
o
CD
OO
CXI
o
0
CO
^j-
0
0
CXI
co
^—
o
o
^j-
o
LO
o
1
1 — *
CXI
o
1
1 —
CD
1
oo
to
s_
o
-f->
fO
1 —
3
fO
S-
O3
S- to
d) S-
N d)
•r- >•>
r— S-
•r- Q
+J
s- -a
d) C
CXI
o
o
CO
[\
LO
o
LO
CO
o
LO
o
o
00
o
1
en
0
o
1
oo
0
1
oo
1 1
o
CO
,
d)
dJ
-f_)
oo
"^
c~
fT5
c
o
S-
o
LO
o
o
1 —
CXI
o
CXI
"~^
o
CD
CXI
OO
"~-
LO
O
1
en
0
o
i
CO
o
i
CO
0
n3
.
O
d)
i —
LU
O
CXI
O
0
LO
LO
0
0
I--.
CXI
o
o
LO
o
oo
o
1
co
0
o
i
CO
CD
1
OO
en
f—
•i —
S-
d)
4_J
c~
•i —
to
LO
OO
0
O
o
CO
CXI
o
o
CXI
o
o
LO
o
LO
LO
o
o
1
oo
o
0
1
«^-
0
1
oo
to
ro
1 —
o
a.
^
>^
S-
•o
c~
^3
o
u_
c
0
>-
CO
o
CD
0
o
00
0
o
CO
1 — •
o
00
co
o
o
CXI
r~-
CXJ
o
1
en
o
o
1
CXI
o
1
"
*
LU
k
CQ
. ,
, —
1 —
•r~
s
CL
,
Z5
D-
s-
-(-> d)
4— i —
fD -i-
S- O
^ CD
o
LO
o
o
LO
^j-
0
o
LO
oo
0
CXI
CO
0
1
LD
1 —
0
1
LO
CD
1
00
to
dJ
o
^_
^
o
to
.
o
01
oo
.JJ
c
rO
1 —
Q.
a>
E
•1 —
1
CD
LO
o
o
LO
•=3-
0
o
LO
CO
o
CXI
CO
CXI
CD
1
LO
0
1
LO
0
1
CO
CD
,—
0
LO
C^J-
o
LO
CO
o
LO
r-
0
0
10
o
o
1
1 —
o
o
1
1 —
o
1
LO
to
s_
o
.)_-)
(O
S-
d>
c
•r—
0
c
1— 1
,—
rO
Q.
.f—
o
• r—
c
3
2"
0
1 —
o
LO
^j-
o
LO
oo
0
LO
J-^.
0
0
LO
o
CXI
0
1
1 —
o
0
1
, —
0
1
-42-
-------
CO
CM
LO
o
I
0
o o
o o
o o o o
o o o o
o o o o
I
LO
o
CM CM
o o
o o
o o
o o
LO
o
o
o
o o
oooo
o o o o
o o o o
UJ LO
CD
CtL
UJ
co co
o o
o o
en en
CM CM
o o
o o
o o
oooo
0000
CM
o
LO LO LO LO
o o o o
o o o o
o o o o
CO
CM CM
O O
O O
O O
LO
LO
O
O O
oooo
LO LD LO LO
oooo
o o o o
o o o o
Cn
oooo
oooo
<
d.
o o
o o
CO CO
•* >*•
o o
o
CM
CM
O O
LO LO LO LO
en en en en
o o o
o o
LD LD LD LD
oooo
o o
o o
CO CO
o
LO
CM
oooo
en en en en
oooo
oooo
oooo
LO LO LO LO
oooo
LlJ
M
CO
>-
OL
UJ
O
O
O
O
co
0
1
CM
0
O
O
1
CM
O
1
CM
O
o
o
CM
o
0
o
1
o
8
CM
0
o
o
1
CO
o
CM
O
o
0
1
o
1
LD
o
0
i
o
CM
O
1
vo
0
o
I
o
1
0
1
*d-
o
o
o
1
o
1
LO
o
o
1
o
CM
o
1
LO
o
o
1
o
CO
o
1
LO
o
o
1
o
1
o
1
CM
o
o
1
o
CM
o
CM
o
o
1
CM
o
1
00
o
1
CM
0
o
1
o
CM
CM
O
O
CM
O
O
1
CM
O
o
CM
O
1
en
o
i
CM
0
o
i
o
i —
i
CM
0
O
1
CM
0
1
co
i
CM
0
O
1
o
1
•=£
O
i—i
CtL
Q- O
t—i
Q-
(O
O
(J
cu
^t
o
4->
co
(O
o
o
cu
c
o
o
>,
o
tc
co
CO
0-
a:
CL
O
o
O
co
o
cu
UJ
to
o
o
>
O-
i—'
rC
•i—
S-
l/l
-o
c
tQ
o
o
S-
QJ
.*:
o
4->
co
-43-
-------
oo
ro
LT>
O
I
CD
O CD O CD O
000000
1
i_n
o
C> O CD O O
o o o o o o
C/O
LO
I
o o o o o
O O O O CD O
O
a;
<=c
Q.
I
LO
ooooo oooooo
CO
I
O O
en cr>
O O O
CT> CTi CT)
OOOOOO
M
I—I
oo
i
CO
ooooo
oooooo
cr> en o^ o> cr» CTI
O
I—
z
UJ
Q
o
CJ
i— CM CO i— CO
O CD O O O
I I I I I
^0 t«o **o r~*- CD
O O O U3 i—
ooooo
I
CM CO
o o
I I
OJ CO
I
I
I
I
I
oo c\j oo c\j cxj
o
i
o o
I
I
un
ooooo
ooooo
i
CM
i
CM CM
i
i
O O O O
CM CM
I
I
I
I
I
I
LT3
O
o
I
CM
O
I
a:
<
D-
un
^.
00
Di
D-
O
(/I
a>
CD
oo
O-
o
O
00
-44-
-------
o
I
CM
O
o • o
o
un CM
i
tn
o
CO
o
oo
o o
LO
•vf cr> CM
01 Ln CM
CO
Ln
i
oo 10
<-o CM
UD r—
o o
CM
kO
cn
Ln
en
UJ
M
oo
CO
CO
o
r^ oo
cn oo
CM
CO
CTi
o;
<
O-
oo
I
cncn
oo
cocriLn
crioor^-
en
en
i
oo
t--
cn
en co
en en
o
en
oo
en oo
en en
o
en
en
en
en
en
UJ
1
1 — 1
U_
1
0
0
I |
I_L_
UJ
• •
Q
i — i
1 1 1
1 1 1
o
> 0
UJ O
f~~\ o
o;
o
o
0
CM
o
o
OO
o
o
r^ oo o
o o .—
o o o
o
CM
0
ID
O
CtL
<
Q.
10
in
•^.
oo
CO
Q-
Q
O
UJ
Q
o
o
c
CO
cr
LU
o
CL>
•I—
o
CD
cn
S-
cu
o
oo
-!->
0)
o
CO
o
£
CO
-o
CO
E
CJ
c
CO
o
•I—
M-
O)
o
l+-
M-
co
O)
cn -a
•r- CO
S-
o
o
CO
o
o
cn
3
c
CO
CJ
CO CO
-C
cn -o
•r- CO
o
4J
-45-
-------
TEC JICAL REPORT DATA
(Please read lum-u^ lions on the reierse be/ore completing/
1. REPORT NO.
EPA-600/4-77-017
4. TITLE AND SUBTITLE
REGIONAL AIR POLLUTION STUDY
Sulfur Compounds and Particulate Size Distribution
Inventory
5. REPORT DATE
April 1977
6. PERFORMING ORGANIZATION CODE
3. RECIPIENT'S \CCESSIOf*NO.
7 AUTHOR(S)
Fred Littman, Robert W. Griscom, and Harry Wang
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Air Monitoring Center
Rockwell International
11640 Administration Drive
Creve Coeur, MO 63141
10 PROGRAM ELEMENT NO.
1AA603
11. CONTRACT/GRANT NO.
68-02-1081
Task Order 56
12 SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
- RTP, NC
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
In conjunction with the Regional Air Pollution Study being conducted in
the St. Louis Air Quality Control Region (AQCR), a methodology for estimating the
amount of sulfur trioxide (SO ) emitted by combustion sources was developed. It
is based on SO /SO ratios determined both experimentally and from literature
surveys. Th~ most likely value appears to be 1.85% of the SO,, emissions. On
fl-~ most l:kely value appears to be 1.85% of the SO emissions.
this basis, about 22,000 tons of SO are emitted yearly from combustion sources
A fine particle size inventory for the area was also developed. The inventory
gives a breakdown of particulate emissions in the range of 7 to .01 microns,
based on production rates and collection efficiencies for point sources in the
St. Louis AQCR. The information on the SO /SO ratios and the particle size
breakdown is stored in the RAPS Data Handling System.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
*Air pollution
*Sulfur tiroxide
*Particle size distribution
*Estimates
*Environmental surveys
St. Louis, MO
13B
07B
05J
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
54
20. SECURITY CLASS (This page!
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
46
------- |